Title :
Design of two-channel IIR linear phase PR filter banks
Author :
Zhang, Xi ; Yoshikawa, Toshinori
Author_Institution :
Dept. of Electr. Eng., Nagaoka Univ. of Technol., Niigata, Japan
Abstract :
This paper proposes a novel method for designing two-channel biorthogonal perfect reconstruction (PR) filter banks with exact linear phase using IIR filters. Since the structurally PR implementation is adopted, the proposed filter banks are guaranteed to be PR even when all filter coefficients are quantized. From the viewpoint of wavelets, design of IIR linear phase filter banks with an additional flatness condition is considered. The proposed design method is based on the formulation of a generalized eigenvalue problem using the Remez exchange algorithm. Therefore, the filter coefficients can be obtained by solving the eigenvalue problem, and the optimal solution with an equiripple magnitude response is easily obtained through a few iterations. The proposed procedure is computationally efficient, and the flatness condition can be arbitrarily specified
Keywords :
IIR filters; digital filters; eigenvalues and eigenfunctions; equiripple filters; filtering theory; frequency response; linear phase filters; signal reconstruction; wavelet transforms; IIR filter banks; PR filter banks; Remez exchange algorithm; biorthogonal filter banks; equiripple magnitude response; filter coefficients; flatness condition; generalized eigenvalue problem; iterations; linear phase filter banks; optimal solution; perfect reconstruction filter banks; two-channel filter banks; wavelets; Biomedical signal processing; Channel bank filters; Design methodology; Ear; Eigenvalues and eigenfunctions; Filter bank; Finite impulse response filter; IIR filters; Image reconstruction; Signal processing algorithms;
Conference_Titel :
Signal Processing Proceedings, 1998. ICSP '98. 1998 Fourth International Conference on
Conference_Location :
Beijing
Print_ISBN :
0-7803-4325-5
DOI :
10.1109/ICOSP.1998.770137